The Jun Kinase/Stress-activated Protein Kinase Pathway Functions to Regulate DNA Repair and Inhibition of the Pathway Sensitizes Tumor Cells to Cisplatin*

We have studied the role of Jun/stress-activated protein kinase (JNK/SAPK) pathway in DNA repair and cisplatin resistance in T98G glioblastoma cells. JUN/SAPK is activated by DNA damage and phosphorylates serines 63 and 73 in the N-terminal domain of c-Jun, which is known to increase its transactivation properties. We show that treatment of T98G glioblastoma cells with cisplatin but not the transplatin isomer activates JNK/SAPK about 10-fold. T98G cells, which are highly resistent to cisplatin (IC50 = 140 ± 13 μm), modified to express a nonphosphorylatable dominant negative c-Jun (termed dnJun) exhibit decreased viability following treatment with cisplatin, but not transplatin, in proportion (r Pearson = 0.98) to the level of dnJun expressed leading to a 7-fold decreased IC50. Similar effects are observed in U87 cells, PC-3 cells, and MCF-7 cells, as well as in T98G cells modified to express TAM-67, a known inhibitor of c-Jun function. In contrast, no sensitization effect was observed in cells modified to express wild-type c-Jun. Furthermore, through quantitative polymerase chain reaction-stop assays, we show that dnJun expressing cells were inhibited in repair of cisplatin adducts (p = 0.55), whereas repair is readily detectable (p = 0.003) in parental cells. These observations indicate that the JNK/SAPK pathway is activated by cisplatin-induced DNA damage and that this response is required for DNA repair and viability following cisplatin treatment. Regulation of DNA repair following genotoxic stress may be a normal physiological role of the JNK/SAPK pathway.

JNK/SAPK 1 is part of a kinase cascade that phosphorylates the transcription factor c-Jun at serine residues 63 and 73 (1)(2)(3)(4)(5)(6)(7)(8)(9). Phosphorylation of c-Jun at these sites greatly enhances the transactivation potential of the AP-1 binding sites (1)(2)(3)(4) and AP-1 regulated genes in vivo (5,11,12), and there is evidence suggesting roles for c-Jun phosphorylation in cellular transformation (1,2), inflammation (14), and apoptosis (15). The JNK/SAPK pathway is strongly stimulated in a dose-dependent manner by various DNA damaging treatments, including UV-C (5,(7)(8), ionizing radiation (16), and alkylating agents such as N-methyl-NЈ-nitro-N-nitrosoguanidine (MNNG) (5), methylmethanesulfonate (MMS) (11), 1-␤-D-arabinofuranosylcytosine (Ara-C) (17), and hydrogen peroxide (18). These observations suggest that the JNK/SAPK pathway may mediate a physiological response to DNA damage such as induction of one or more DNA repair enzymes. Here we provide evidence that the chemotherapeutic agent cisplatin, which damages DNA through the formation of bifunctional platinum adducts, but not transplatin, which does not damage DNA (19,20), activates JNK/SAPK up to 10-fold in a dose-dependent manner. Furthermore, inhibition of this pathway in cells modified by expression of a nonphosphorylatable dominant negative mutant of c-Jun, dnJun, blocks DNA repair as judged by quantitative PCR and markedly decreases viability following treatment with cisplatin but not transplatin. Thus, JNK/SAPK is activated by cisplatin-induced DNA damage and is required for DNA repair and survival following cisplatin treatment.

EXPERIMENTAL PROCEDURES
Cells-Culture conditions and all cell lines and plasmids used here were developed using standard methods as described previously (22,23). The expression of c-Jun and dnJun was quantitated using the methods (24) and antibodies previously characterized (24).
PCR-The PCR-stop assay (28) was used to quantitate cisplatin-DNA adduct formation and subsequent repair. The assay is based on the observation that the efficiency of amplification of cisplatin-treated DNA is inversely proportional to the degree of platination. Genomic DNA was isolated immediately or 6 h after treatment of cells for 1 h 15 min with varying amounts of cisplatin and amplified quantitatively using 32 P-end-labeled primers, giving rise to a 2.7-kb and a nested 0.15-kb fragment of the hyopxanthine phosphoribosyl transferase gene. The 5Ј and 3Ј primers were TGGGATTACACGTGTGAACCAACC and GATCCACAGTCTGCCTGAGTCACT, respectively, with a 5Ј nested primer of CCTAGAAAGCACATGGAGAGCTAG. The 0.15-kb segment of genomic DNA sustains undetectable levels of DNA damage under our conditions and serves as an internal PCR control and the basis for normalization of the amount of amplification of the 2.7-kb fragment. The number of lesions/2.7-kb fragment (i.e. Fig. 4) is calculated as 1 Ϫ (cpm damaged DNA/cpm undamaged DNA) (8).
JNK Assay-JNK assays were carried out exactly as described previously (7).
Cytotoxicity-Viability (29) was assessed by the addition of cisplatin or transplatin for 1 h one day after seeding test cells into 96-well plates followed by a change of medium to fresh medium and determination of surviving cells 5 days later by addition of MTS for 1 h and determination of A 590 nm of the dissolved formazan product as described by the  This paper is available on line at http://www-jbc.stanford.edu/jbc/ manufacturer (Promega). All results were carried out in quadruplicate, and viability is expressed as the ratio of the amount of viable cells following cisplatin or transplatin treatment to that of the same cells without treatment.

Activation of JNK/SAPK by Cisplatin Requires DNA Adduct
Formation-It is known that cisplatin but not transplatin forms covalent covalent cross-links between the N 7 position of adjacent guanine or adjacent adenine-guanine residues (19,20). We find that the JNK activity of T98G cells is elevated in a dose-dependent manner up to 10-fold following a 1-h exposure to cisplatin but not to transplatin (Fig. 1A). As a positive control of the effects of a DNA-damaging agent, we examined the response of JNK of T98G cells to UV-C irradiation (Fig. 1B) and observed a similar dose-response relationship. Cisplatinspecific responses have been observed in other cell lines from tumor types that are commonly refractory to cisplatin treatment such as the human nonsmall cell lung carcinoma lines A549 (data not shown) and M103 (Fig. 1C). Moreover, 1 h after treatment with cisplatin, but not transplatin, JNK activity of T98G cells or lung carcinoma cells M103 remains elevated, suggesting that treatment with cisplatin leads to a prolonged response. These results indicate that only the DNA-damaging cisplatin isomer activates JNK activity.
Dominant Negative c-Jun Sensitizes Tumor Cells to Cisplatin but Not Transplatin-We developed clonal lines of human T98G glioblastoma cells, which stably express a dominant negative inhibitor (1, 2) of the JNK/SAPK pathway, dnJun. Expression of dnJun has no effect on either basal AP-1 activity (1, 2) or on the enzyme activity of JNK (data not shown) but does inhibit phosphorylation-dependent activation of transcription (1,2,10,12). The effect of cisplatin treatment on the viability of representative clonal lines of the dnJun-expressing T98G cells is compared with that of an empty vector control line, T98GLHCX, in Fig (Table I). In contrast, the dnJun expressing cells exhibit an IC 50 as low as 21 M ( Fig. 2A) or over 7-fold more sensitive to cisplatin than the control cells (Table I). Replicate experiments using additional clones that exhibit varying amounts of steady state dnJun indicate the sensitization to cisplatin is proportional (r Pearson ϭ 0.98) to the amount of dnJun expressed (Fig.  2B). Transplatin has no discernible effect at concentrations where the viability following treatment with cisplatin is less than 25% (Fig. 3B) and in extended titrations no significant effect at 250 M, indicating that the requirements for sensitization by dnJun depends upon the stereospecific DNA-binding properties of cisplatin, similar to the conditions for the activation of JNK.
Expression of wild-type c-Jun does not mimic dnJun-expressing cells (Fig. 2A). In fact, the viability of these cells when treated with cisplatin is somewhat increased relative to parental or empty vector control cells for all viability determinations in the range 20 -60 M cisplatin, suggesting that increased JNK substrate augmented viability following treatment with cisplatin ( Fig. 2A). Thus, the sensitization to cisplatin observed for the dnJun-expressing cells appears to correlate with interference in the role of activated c-Jun.
Generality-We have tested the generality of the sensitizing properties of dnJun in PC3 prostate carcinoma cells modified to express dnJun under the control of an inducible truncated metallothionein promoter as described previously (21). Viability studies show that parental or empty vector control cells are largely insensitive to cisplatin at concentrations Յ60 M (Fig.  3A, circles). Extended titrations revealed IC 50 values of 109 and 156 M for the parental and empty vector control cells, respectively (Table I). However, for PC3 cells that stably express pMTdnJun, the IC 50 value is markedly reduced (Fig. 3A, open  diamonds). Further, induction of maximum expression of dnJun by the addition of zinc acetate leads to greatly increased cytotoxicity with an IC 50 of 16 M (Fig. 3B) or 7.24 -9.8-fold more sensitive to cisplatin than control cells (Table I). The addition of zinc acetate alone has no effect on the viability of parental or empty vector control cells (Fig. 3A, filled circles). Thus, the results observed following induction of expression of dnJun by a single clonal line confirm the results of Fig. 2B that sensitization to cisplatin is dependent upon the expression of dnJun.
As a further control, we examined PC3 prostate carcinoma cells modified to express a zinc-inducible c-Jun derivative, TAM-67, a well characterized transdominant negative inhibi-

FIG. 2. dnJun sensitizes T98G cells to cisplatin. A, viability assay of empty vector control cells (q), wild-type c-Jun expressing cells (f), and clonal dnJun-expressing cells (ࡗ). B, dose-response curve of the IC 50
cisplatin of dnJun-expressing clones of T98G cells versus total immunoreactive c-Jun (c-Jun ϩ dnJun). Total immunoreactive Jun (c-Jun ϩ dnJun) was determined by sequential immunoprecipitation of 35 S-labeled cells using specific Jun B, Jun D, followed by pan-Jun antiserum as described previously (24). Values above 8.5 are taken as expression of dnJun. FIG. 1. Cisplatin is a stereo-specific activator of JNK. A, T98G human glioblastoma cells were stored overnight in serum-free medium, plated the next day, and treated by the indicated concentrations of cisplatin or transplatin for 1 h with a 1-h chase period followed by lysis and assay for JNK activity as described (7). Matching wells of cells were harvested and counted (Coulter counter) and used as the basis for sample loading. FBS, fetal bovine serum. B, positive control. T98G cells were exposed to the indicated doses of UV-C band (Stratalinker UV cross-linker 1800) radiation and processed as described for A. C, JNK activity of human lung carcinoma M103 cells following treatment with 200 M cisplatin or transplatin and processing as described for A with the addition of a 1-h chase prior to lysis. tor of AP-1 owing to a deletion of residues 2-122 (21). As with dnJun, induction of TAM-67 in PC3 cells strongly enhances their sensitivity to cisplatin (Fig. 3A). We have determined that these TAM-67 and dnJun are expressed in approximately equal amounts, suggesting that the comparable degree of sensitization for TAM-67 and dnJun (Fig. 3A) is accounted for by interference in the role of phosphorylation-related function of c-Jun.
Similar results have been observed with an additional human glioblastoma line, U87, and an additional epithelial tumor line, MCF-7 (Table I). Clonal dnJun-expressing lines of these cells exhibit 2.6-and 3.8-fold decreased IC 50 values, respectively (Table I). Thus, the sum of results indicate that the JNK/SAPK pathway may have a general role in mediating a functional response to DNA-cisplatin adduct formation. Inhibition of this response sensitizes cells to the cell-killing properties of cisplatin.
Cisplatin Activates and dnJun Inhibits DNA Repair-We assessed the extent of genomic DNA damage and repair following cisplatin treatment using a modified PCR assay (25). For this assay, it has been shown that the degree of inhibition of PCR-catalyzed amplification of DNA purified from cisplatintreated cells is a direct measure of the amount of DNA-cisplatin adduct formation as measured by atomic absorption (25). Thus, this assay provides a direct assessment of the extent of cisplatin-induced DNA damage.
DNA isolated from T98G cells immediately after treatment with 0, 100, or 200 M cisplatin for 1 h exhibit increasing levels of DNA damage (Fig. 4A, circles). However, if a 6-h "recovery" period is introduced prior to the DNA purification, damage is markedly and significantly (p ϭ 0.003) reduced (Fig. 4A, filled  circles). As a positive control for the effects of inhibition of genomic DNA repair, an inhibitor of ADP-ribosylation, 2-aminobenzidine, was added at the time of treatment of the cells with cisplatin (Fig. 4A, squares). Following the 6-h recovery period, DNA damage remained unrepaired, and total DNA damage was substantially increased. Next, we compared the level of DNA damage for T98G cells and dnJun-expressing cells following treatment with cisplatin (Fig. 4B). For the dnJunexpressing T98G cells, 6 h after cisplatin treatment DNA damage remains completely unrepaired for cells treated at either 100 or 200 M cisplatin (p Ͼ 0.53). All the results summarized here (Fig. 4, A and B) are the averages of three independent assays, which confirms the reliability of this observation. The sum of results, therefore, strongly indicates that expression of dnJun by T98G cells largely abolishes DNA repair following exposure of the cells to cisplatin. DISCUSSION These studies show that the JNK/SAPK pathway is activated by cisplatin-induced DNA damage and is required for DNA repair and viability following cisplatin treatment. T98G glioblastoma cells modified to express a nonphosphorylatable dom-  inant negative inhibitor of c-Jun, dnJun, fail to repair cisplatin adducts and are sensitized to the cytotoxic effects of cisplatin under conditions that have little or no effect on parental and control lines. In contrast, cell lines modified to overexpress wild-type c-Jun are resistant to cisplatin, an observation that rules out that possibility that the sensitization effect of dnJun is mediated by one or more of the domains it shares with wild-type c-Jun. Moreover, sensitization to cisplatin by dnJun is exhibited by several cell lines of varying origins. Sensitization to cisplatin is also observed in PC-3 prostate carcinoma cells modified to express TAM-67, a known dominant negative inhibitor of AP-1 (21). Because the degree of protein expression and sensitization is similar for TAM-67 and for dnJun, we conclude that most of the sensitization effects we observe are accounted for by inhibition of the phosphorylation-related functions of Jun. Two major types of DNA regulatory elements that respond to the phosphorylation state of c-Jun include classic AP-1 sites and ATF/CREB sites. Classic AP-1 sites consisting of a 7-base pair consensus motif, T(G/T)A(C/G)TCA, bind to AP-1 complexes consisting of heterodimers of members of the Fos and Jun families and to Jun-Jun homodimers (9, 13-15, 21, 22, 26). ATF/CREB sites consisting of an 8-base pair consensus motif, T(G/A)CGTCA, bind to c-Jun/ATF2 heterodimers. Indeed, because JNK phosphorylates ATF2 as well as c-Jun and promotes complex formation and binding to ATF/CREB sites, these sites are likely to be major targets of JNK-mediated regulation (9,12,14,15). Several enzymes known to be involved in repair of DNA-cisplatin adducts and implicated in cisplatin resistance (20) contain ATF/CREB sites in their promoters including DNA polymerase ␤ (27, 28), topoisomerase I (30,31), and proliferating cell nuclear antigen, an accessory protein of DNA polymerase delta (32,33). Moreover, transcription of these genes is known to be activated through the ATF/CREB sites upon stimulation by genotoxic agents (27)(28)(29)(30)(31)(32)(33). Thus, the inhibition of induction of any or all of these activities could account for the inhibitory effects of dnJun on DNA repair and the resultant increase in cisplatin sensitivity. In view of the common regulatory mechanism involving ATF/CREB sites, a concerted induction of genes with a related function, DNA repair, is suggested. The sum of results indicate, therefore, that a potential physiological role for the strong activation of the JNK/SAPK pathway following DNA damage may be to mediate DNA repair by enhancing transaction of DNA repair enzymes.